ORIGINAL RESEARCH
published: 18 November 2016
doi: 10.3389/fmars.2016.00241
Habitat Differences in Deep-Sea
Megafaunal Communities off New
Zealand: Implications for
Vulnerability to Anthropogenic
Disturbance and Management
Ashley A. Rowden *, Daniel Leduc, Malcolm R. Clark and David A. Bowden
National Institute of Water and Atmospheric Research, Wellington, New Zealand
Edited by:
Anthony Grehan,
National University of Ireland, Galway,
Ireland
Reviewed by:
Tina Molodtsova,
Shirshov Institute of Oceanology,
Russia
Rebecca E. Ross,
Plymouth University, UK
*Correspondence:
Ashley A. Rowden
[email protected]
Specialty section:
This article was submitted to
Deep-Sea Environments and Ecology,
a section of the journal
Frontiers in Marine Science
Received: 04 August 2016
Accepted: 04 November 2016
Published: 18 November 2016
Citation:
Rowden AA, Leduc D, Clark MR and
Bowden DA (2016) Habitat
Differences in Deep-Sea Megafaunal
Communities off New Zealand:
Implications for Vulnerability to
Anthropogenic Disturbance and
Management. Front. Mar. Sci. 3:241.
doi: 10.3389/fmars.2016.00241
Research on benthic communities in the deep sea has focused largely on habitats in
isolation, with few studies considering multiple habitats simultaneously in a comparable
manner. The present study aimed to determine the structural differences in benthic
communities of continental slope, seamount, canyon, vent, and seep habitats, and
assess their relative vulnerabilities to disturbance from bottom trawling and potential
seabed mining. Megafaunal invertebrate communities of these habitats were sampled
in two regions off New Zealand, in four depth strata between 700 and 1500 m, using
an epibenthic sled and a beam trawl. Patterns of community and trophic structure, and
the potential influence of environmental variables, were determined using multivariate
analyses. The difference in community structure between regions was greater than
among habitats and depth strata. Levels of food availability may explain regional
differences in community structure, although some influence of fishing disturbance is
also possible. Differences in community and trophic structure were most pronounced
between the chemosynthetic vent and seep habitats, and other habitats. Differences
among these other habitats within a region were inconsistent, except that canyon and
slope communities always differed from each other. Community and trophic structural
patterns were partly explained by the environmental differences observed among
habitats. The relative vulnerabilities of benthic communities to human disturbance in the
two regions were determined based on patterns of abundance and feeding mode of
the megafauna. Communities of vent and seep habitats were assessed to be more
vulnerable to disturbance than those of the other habitats based on a number of
habitat-related attributes. However, the relative vulnerability of megafaunal communities
at slope, canyon, and seamount habitats could not confidently be assessed on a habitat
basis alone. The results of the present study have implications for how regional and
habitat differences in benthic communities are incorporated into spatial management
options for the deep sea.
Keywords: megafauna, benthic, community, habitat, human disturbance, vulnerability
Frontiers in Marine Science | www.frontiersin.org
1
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
INTRODUCTION
structure, this may indicate that they are relatively unconnected
(Kelly et al., 2010). Consequently they may be vulnerable to
anthropogenic impacts because they have limited ability to
recover via dispersal if disturbed at a habitat-level scale. Faunal
dissimilarity among habitats could also indicate that a habitat’s
environmental characteristics are distinct enough to allow the
development of specific communities (Kelly et al., 2010). Some
functional or trophic groups within megafaunal communities
are particularly sensitive to disturbance, which may result in
the community of deep-sea habitats being more vulnerable to
disturbance than others if the proportion of these taxa is relatively
high or of particular importance for the structure and function
of the community. The sensitivity of such organisms may be
expressed by a variety of structural, positional, or ecological traits
(e.g., Tyler-Walters et al., 2009). Feeding mode is an important
trait; for example, megafaunal filter- and suspension-feeders,
which are abundant in hydro-dynamically active parts of canyons
and seamounts (Genin et al., 1986; Duineveld et al., 2001), are
vulnerable to smothering by the gravity flows caused by bottom
trawling along canyon edges (Palanques et al., 2006; Martín et al.,
2014) or physical impacts from trawling that focuses on the
summits of seamounts (Clark and O’Driscoll, 2003; Clark and
Rowden, 2009).
In New Zealand, deep-sea fisheries are economically valuable
but can impact habitats with vulnerable benthic communities
(e.g., seamounts: Clark and Rowden, 2009; Williams et al., 2010b;
canyons: De Leo et al., 2010; seeps: Baco et al., 2010; Bowden
et al., 2013), and the fauna of the open slope (e.g., Cryer
et al., 2002). Consumers are becoming more aware of the issues
surrounding the sustainability of fisheries, thus research that
can inform approaches to balance fisheries exploitation with
conservation of vulnerable deep-sea communities are of interest
to resource and environmental managers. In New Zealand, a
limitation of a previous comparison of fauna among different
deep-sea habitats with potential vulnerable communities is
that it examined one taxonomic group and depended on the
compilation and manipulation of data obtained by different
surveys using varying methodologies (Rowden et al., 2010b).
Therefore, advances in understanding the similarity of fauna
among different deep-sea habitats, and thus their relative
vulnerabilities to disturbance, need to come from studies that
examine a range of communities across different habitat types
using comparable sampling methods. The New Zealand EEZ
is an ideal location to conduct such a study because a diverse
array of deep-sea habitats occur in close geographical proximity
(Gordon et al., 2010), enabling direct comparisons between
habitats without the confounding factor of spatial separation.
The present study examined the following hypotheses: (1) the
benthic communities of deep-sea habitats such as seamounts,
vents, canyons, and seeps are structurally and functionally
different from each other and from those found on the
continental slope; (2) the communities of these habitats are
more vulnerable to human disturbance than are those of
continental slope environments; and (3) the effects of fishing
and potentially other human activities are more severe for these
communities than for those on continental slopes. Data collected
to address these hypotheses will ultimately inform ecological risk
Research on benthic communities in the deep sea has focused
largely on habitats in isolation (e.g., slope, Palma et al., 2005;
vents, Desbruyères et al., 2006; canyons, Schlacher et al.,
2007; seeps Levin and Mendoza, 2007; seamounts, Lundsten
et al., 2009), with few studies considering even two habitats
simultaneously in a comparable manner (canyons and slopes:
Vetter and Dayton, 1998; e.g., seamounts and slopes: O’Hara,
2007; McClain et al., 2009; slope and seeps, Zeppilli et al.,
2012; De Leo et al., 2014; abyssal hills and plain, Durden et al.,
2015; vents and seeps, Nakajima et al., 2015). Only relatively
recently, has the need to consider biodiversity patterns and
the potential linkages between multiple deep-sea habitats been
identified and addressed (e.g., Vanreusel et al., 2010; Kelly
et al., 2010; Sevastou et al., 2013). The need to understand
connectivity between habitats arises not only from scientific
interest in obtaining greater ecological understanding of the deep
sea, but also from social pressures to generate information for the
effective management of the growing industrial use of the deep
sea (Ramirez-Llodra et al., 2011).
Megafauna (animals > ∼3–5 cm in dimension) are
a commonly studied component of benthic invertebrate
communities in the deep sea, sampled either directly by sleds
or trawls, or indirectly by photographic methods. Photographic
sampling focuses on the mega-epifauna (seafloor surface
dwelling fauna), while direct sampling can include infauna
(sediment dwelling) as well as epifauna. Studies of megafaunal
communities along continental margins (sampled by sleds and
trawls—the focus of this study) have identified the influence
of habitat heterogeneity, with variations in environmental
parameters such as oxygen, temperature, overlying productivity
etc., affecting broad-scale patterns of community structure
(species composition and abundance) (e.g., Sellanes et al.,
2010; Williams et al., 2010a). Nested within these regional
environmental variations along a margin are a range of more
localized topographic habitats such as canyons, seamounts,
banks, and ridges as well as the open continental slope. Cold
seep and hydrothermal habitats also occur on or adjacent to
some continental margins (Menot et al., 2010). Studies have
shown that these feature-scales of habitat heterogeneity influence
the biodiversity patterns (abundance, biomass and diversity)
of megafauna on continental margins due to differences in
various local environmental factors (e.g., canyon versus open
slope—Sardà et al., 1994; Ramírez-Llodra et al., 2008; Vetter
et al., 2010; seamount versus slope—Rowden et al., 2010a;
Tecchio et al., 2013). However, few studies have determined and
reported the level of community similarity (species turnover or
β-diversity) among the megafaunal communities that inhabit
these deep-sea habitats (Cartes et al., 1994; Ramirez-Llodra et al.,
2010; Rowden et al., 2010b), and only one to our knowledge
has used community similarity to explicitly assess the degree of
connectivity between habitats (Kelly et al., 2010—which included
megafauna), and none to assess their potential vulnerability to
anthropogenic disturbance.
If benthic communities of different habitats in the deep sea
are generally dissimilar to one another in terms of community
Frontiers in Marine Science | www.frontiersin.org
2
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
Sampling and Data Acquisition
assessment, helping environmental and resource managers to
assess and mitigate the threats to vulnerable deep-sea habitats and
their communities whilst maintaining sustainable exploitation of
resources.
The current study focuses on megafaunal communities of
deep-sea habitats sampled by benthic sled and beam trawl.
This study forms part of a complementary set of investigations
conducted under a larger programme. Results for mega-epifauna
sampled by towed video camera, and macro-infauna and
meiofauna sampled by multicorer and box-corer, have been
reported separately (Bowden et al., 2016; Leduc et al., 2016; Rosli
et al., 2016).
Sampling of the slope, canyon and seamount habitat sites was
conducted during a single survey of each region. Sites on the
Hikurangi Margin were sampled from RV Tangaroa in April 2010
(TAN1004), and the Bay of Plenty was sampled in April/May
2012 (TAN1206). Data for seep and vent habitat sites were almost
wholly obtained from separate and independent surveys. Data for
seep sites on the Hikurangi Margin were obtained from a survey
in 2006 (TAN0616) (Baco et al., 2010). One vent habitat site in the
Bay of Plenty was sampled during the TAN1206 survey, with data
for others obtained from surveys in 2001 and 2011 (TAN0107 and
TAN1104, respectively).
During the two main surveys, benthic communities at each
of the sites for slope, canyon, and seamount habitats were
sampled at four depth strata (700, 1000, 1200, and 1500 m).
This stratification allowed for the influence of depth effects
to be accounted for in statistical analyses, thereby providing
a more robust evaluation of any habitat effect. High quality
multibeam echosounder data were already available for most of
the planned sampling sites but where it was necessary, multibeam
transects (using the RV Tangaroa’s Kongsberg EM302) were run
before sampling commenced. At each sampling site, a towed
camera was deployed first to determine substratum type and to
record photographic transects before disturbance from physical
sampling. Deployment of the sampling gear was planned with
direct reference to multibeam maps and observations from the
video transects. For predominantly soft substratum, a beam trawl
was used to sample megafauna, while on predominantly hard or
rough mixed substrata where there was risk of damage to the
beam trawl, a sled was deployed. Both gears were towed along
the same line as the camera transect.
The beam trawl (4 m width, 25 mm cod-end mesh) is a design
based on the “Honfleur” fish trawl (Guennegan and Martin,
1985). It was modified using a ground rope with small-diameter
(180 mm diameter) rubber disk rollers spaced by 75 mm cookies.
This modification was done to enable the trawl to be used on
rougher seabed and to reduce the tendency of the net to dig
in on muddy substrates. The sled (1 m width, 25 mm cod-end
mesh) is of sturdy construction and designed to operate on rough
ground, such as found on seamounts (Clark and Stewart, 2016).
Both the trawl and the sled were deployed for 10-15 min bottom
time.
See Supplementary Table 1 for station details.
METHODS
Study Area
In order to make robust comparisons of benthic communities
from different deep-sea habitats, the geographic distance between
habitats should be minimized and sampling should occur within
a uniform environmental setting. There is no single localized area
off New Zealand where all continental margin seabed habitats cooccur, however, so two regions were selected that are each missing
only a single, and different, chemosynthetic habitat. These
regions are the Hikurangi Margin (slope, canyons, seamounts,
and seeps), and the Bay of Plenty (slope, canyons, seamounts, and
vents). Sampling of two regions allows a more general evaluation
of dissimilarity in benthic community structure observed among
the different habitats. The two regions also cover a range of
human activities. The Hikurangi Margin has significant fisheries,
including for hoki (Macruronus novaezelandiae) and orange
roughy (Hoplostethus altanticus) (Clark, 1995), and may be
subject to deep-seabed mining for gas hydrates in the future
(Fohrmann and Pecher, 2012). The Bay of Plenty also has
some deep-sea trawl fisheries, including for orange roughy and
alfonsino (Beryx decadactylus) (Clark and O’Driscoll, 2003), and
is of deep-sea mining interest for seafloor massive sulfide deposits
(Boschen et al., 2013).
The Hikurangi Margin is located on the south-eastern coast
of New Zealand’s North Island (Figure 1). The slope is incised
with a number of canyons and there are some physically isolated
bank and seamount-like features that occur between the canyons
along the general slope (Mountjoy et al., 2009). On some of the
broader banks there are active methane seeps (Greinert et al.,
2010). The Bay of Plenty is located off the north-eastern coast
of New Zealand’s North Island (Figure 1). The slope of this
region is also incised with canyons, while the seamounts are
part of the Kermadec volcanic arc that trends to the north-east
(Wysoczanski and Clark, 2012). Some of the seamounts of the
Kermadec volcanic arc have active hydrothermal vents (de Ronde
et al., 2007). Three sites for each of the habitats common to
both regions (slope, canyon, seamount) were selected, as much
as possible of similar size, morphology, and close together. Seep
habitat sites on the Hikurangi Margin were constrained to the
Opouawe Bank in the middle of the study region, while vent
habitat sites in the Bay of Plenty were either co-located with the
study seamount sites or on seamounts immediately further north
on the Kermadec arc (Figure 1).
Frontiers in Marine Science | www.frontiersin.org
Sample and Data Treatment
The whole sample from the beam trawl or sled was photographed
on deck and then transferred to bins for weighing. Depending
on catch volume, either the whole sample or a subsample was
then sorted for fauna on a 5 mm mesh sorting table. All fauna
were preserved appropriately (ethanol or formalin depending on
taxon) except for highly abundant species for which voucher
specimens were retained and the bulk discarded. All preserved
and discarded samples, together with total catch weights, were
recorded in a database and, once ashore, all faunal samples were
lodged in the NIWA Invertebrate Collection, Wellington.
Samples from the two main surveys were distributed
among taxonomists and parataxonomists at NIWA and other
3
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
FIGURE 1 | Sampling sites at habitats in the Bay of Plenty (BOP, top) and Hikurangi Margin (HIK, bottom) study regions, and their relative locations off
New Zealand (insert; crosses on insert shows location of vent sites at Healy and Brothers seamounts on the Kermadec volcanic arc). Isobaths show
200, 500, 1000, and 2000 m depths. Scale bar applies to both regional maps. See Appendix 1 for sample station details.
Samples from other surveys for which data were used in the
community analysis described below were identified by the same
people.
institutions in New Zealand and elsewhere (see Supplementary
Table 2) for identification to species level where possible,
and enumeration (counts of individuals or distinct colonies).
Frontiers in Marine Science | www.frontiersin.org
4
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
total catch. Abundance of organisms in the samples was then
standardized to number of individuals per 100 m2 of seabed, to
take account for the difference in dimensions of the gear (see
above), as well as the difference in tow length between stations.
Preliminary examination of megafaunal datasets showed
the presence of unusually high abundance of Hyalinoecia
longibranchiata (quill worms) in one beam trawl sample in
the Hikurangi Margin region (1630 individuals at station 43,
TAN1004). This single datum exceeds all others by about one
order of magnitude. Ecological theory suggests that species which
are highly clustered should be given less weight in multivariate
community analyses than those that are more evenly distributed
(Clarke et al., 2006a). Thus, quill worm abundance data for this
sample was log(x+1)-transformed to reduce the impact of this
outlier on the analyses.
Several taxa present in the original megafaunal dataset
were removed from final analyses because they were either
ill-defined (and could therefore be confused with other taxa;
e.g., Amphipods, Cnidaria, Holothuroidea), or pelagic (e.g.,
Thaliacea, Scyphozoa).
Stations where the start, end, and mean tow depths
differed from the nominal target stratum depth (four and six
stations in TAN1004 and TAN1206 datasets, respectively, see
Supplementary Table 1) were re-allocated to appropriate strata
prior to conducting analyses to keep the water depth consistent
across habitats. One station with a mean water depth of 495 m
was on the summit of a seamount that was shallower than the
shallowest a priori stratum (700 m) and therefore could not be
re-assigned. Data from this station was included in the analysis
under the assumption that they would not bias the results.
Using information in Rowden et al. (2010a) and the
scientific literature, all taxa were assigned to one of six
feeding modes: suspension-feeder; filter-feeder; deposit-feeder;
predator/scavenger; grazer, or chemoautotrophic. Some of the
feeding mode labels and the taxa assigned to them were revised
slightly from those used by Rowden et al. (2010a) for slope and
seamount benthos in the southwest Pacific (see Supplementary
Table 3 for definitions).
Environmental Data
Substrate data were obtained from video observations of the
camera transects made prior to the deployment of the sled/beam
trawl at each site. Substrate types (bedrock, boulders, cobbles,
pebbles, gravel, sand, mud and the biogenic substrates “coral
rubble” and “shell hash”) were expressed as percentages of the full
transect distance after correcting for any sections of the transect
in which the seabed could not be observed (e.g., because of high
altitude or speed, or seafloor contact) (Bowden et al., 2016). An
additional substrate variable (substrate diversity) was calculated
based on substrate type data using the Shannon-Wiener diversity
index (Etter and Grassle, 1992).
Habitats at the study sites were also characterized using
seafloor morphology derivatives from multibeam data gridded
at 25 × 25 m resolution. The derived topography variables were:
depth, slope (steepest gradient to any neighboring cell), aspect
(direction of slope), curvature (change of slope), plan curvature
(curvature of the surface perpendicular to the slope direction),
and profile curvature (curvature of the surface in the direction
of slope). Additional derivatives were calculated for the standard
deviation of depth, depth range, standard deviation of the slope,
and terrain rugosity based on a 3, 5, 7, and 15 grid cell focal mean.
For the analysis of megafauna data, topography variables based
on the largest grid cell focal mean (i.e., 15 × 15 cells of 25 m or
375 × 375 m) were used, because this spatial scale was closest to
the mean length of sled and beam trawl tows (range: 37–1055 m,
mean = 516 m).
Fishing intensity at each site was obtained from a trawl fishery
database maintained at NIWA for the New Zealand Ministry for
Primary Industries. This database includes position coordinates
of commercial bottom trawls targeting deep water fish species
from July 1980 to March 2011. Fishing intensity was defined as
the sum of all fishing trawls (either defined as lines or points,
the latter when start but not finish location was recorded)
intersecting a 2 km radius circle centered on the mid-point of
each sled/beam trawl tow (to match the precision of reported
fishing trawl trajectories and the length of the sled/beam trawl
and video tows) during the 15-year period prior to sampling at
that site (the maximum period for which records were available
across all sites).
Statistical Approach
Analyses were conducted using routines in the multivariate
software package PRIMER v6 (Clarke and Gorley, 2006).
Analyses of community structure were based on Bray-Curtis
similarity matrices (Clarke et al., 2006b) of square
root-transformed abundance data (Clarke and Warwick,
2001), whereas analyses of environmental variables were based
on Euclidean Distance similarity matrices of normalized data.
Analyses were run in two stages; the first including data only
from the topographic habitats (slope, canyon, and seamount),
the second incorporating vent and seep habitats.
1st Stage Analysis
Megafaunal data from 99 stations were included in analyses of
slope, canyon, and seamount habitats.
The PERMANOVA routine in PRIMER was used to compare
the effects of region, habitat, and depth on benthic community
structure (Anderson et al., 2008). PERMANOVA is a semiparametric, permutation-based routine for analysis of variance
based on any similarity measure (e.g., Bray-Curtis). Analyses
were conducted using the random factor Region (Bay of Plenty,
Hikurangi Margin), and fixed factors Habitat (slope, canyon,
and seamount) and Depth (four strata at ∼700, 1000, 1200, and
1500 m depth). P-values were obtained using 999 permutations.
Because PERMANOVA is sensitive to differences in multivariate
dispersion among groups, the PERMDISP routine in PRIMER
Data Analysis
Pre-Analysis Data Treatment
The abundance values of four (TAN1206) and three (TAN1004)
of the sampling stations were initially based on subsamples
of sled/beam trawl catch due to the large amount of material
obtained. These abundance values were adjusted to take into
account the proportion of material processed relative to the
Frontiers in Marine Science | www.frontiersin.org
5
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
region were compared using PERMANOVA, and visualized
using a MDS ordination plot. SIMPER was used to identify
the taxa contributing most to the among habitat differences
in community structure. PERMANOVA analysis was also
conducted using trophic structure data (untransformed data) as
described above, and the relative representation of trophic groups
among habitats was visualized using pie diagrams. No DistLM
analyses were conducted as part of the 2nd Stage analysis because
it was not possible to obtain matching environmental data for all
samples used from seep and vent habitats.
was used to test for homogeneity of dispersion when significant
factor effects were found (Anderson et al., 2008). The square
root of estimates of components of variation was used for
comparing the amount of variation attributable to the factors in
the multivariate PERMANOVA models (Anderson et al., 2008).
Lack of independence between sites due to geographical
proximity is common in natural communities and presents
problems for the interpretation of ecological patterns (Legendre,
1993). Failure to take into account the spatial component of
ecological variation may affect tests of statistical significance
when investigating relationships between community structure
and environmental parameters such as region, habitat, water
depth (Legendre and Troussellier, 1988). Thus, Latitude and
Longitude were entered as covariates in the PERMANOVA
analyses (Anderson et al., 2008).
Results of the PERMANOVA analysis were visualized using
non-metric multidimensional scaling (MDS) ordinations, with
samples symbol-coded for the factors used in the analysis.
The SIMPER routine in PRIMER was used to identify the
taxa contributing most to differences between regions and among
habitats.
SIMPER was also used to describe the main environmental
variables responsible for habitat differences (using normalized
environmental data). In order to examine more directly the
potential influence of habitat-related variables on community
structure, relationships between megafaunal community
structure and environmental variables including trawling were
investigated using Distance-based Linear Models (DistLMs) in
PERMANOVA+ (Anderson et al., 2008). The DistLM routine is
based on an approach called distance-based redundancy analysis
(dbRDA) first developed by Legendre and Anderson (1999).
Variability in megafaunal community structure was partitioned
in DistLM according to the environmental predictor variables
detailed above, with latitude and longitude fitted first into the
models. Predictor variables that were strongly correlated (R 2 >
0.9) were removed prior to analyses, taking out the variable
most weakly correlated with community structure. Marginal
tests were initially conducted to examine relationships between
each predictor variable and benthic community structure.
Sequential tests were then used to determine which variables
best correlated with megafaunal community structure (using
a step-wise selection procedure with the Akaike information
criterion (AIC)) (Quinn and Keough, 2009). P-values were
obtained using 999 permutations of raw data.
RESULTS
Community Similarity among Slope,
Canyon, and Seamount Habitats
The PERMANOVA analysis showed a significant effect of
Region on community structure but also significant interactions
between Region and Habitat and Region and Depth (Table 1),
indicating that the effects of habitat and depth on community
structure differed between the two regions. Comparing the
estimates of components of variation indicated that the effect
of Region (88.7) was about seven times stronger than both
interaction effects (12.9 and 13.1). Pairwise comparisons (within
Region) showed that, on the Hikurangi Margin, megafaunal
community structure in canyons was significantly different
from that in both slope and seamount habitats (Table 2)
but communities in slope and seamount habitats were not
significantly different. In the Bay of Plenty, community structure
differed significantly among all three habitats. Communities of
most depth strata were significantly different from each other,
with the exception of the 1000 and 1200 m versus the 1500 m
strata on the Hikurangi Margin (Table 2). Despite high 2dimensional stress levels, MDS plots illustrate the patterns of
community similarity underlying the PERMANOVA analysis:
showing a difference in megafaunal community structure
TABLE 1 | Results of PERMANOVA analyses testing for the effects of
Region (Hikurangi Margin versus Bay of Plenty), Habitat (slope, canyon,
and seamount), Depth (700, 1000, 1200, and 1500 m), and their
interactions on megafauna community structure, after accounting for the
effect of spatial covariates (latitude and longitude; results not shown).
Source
Df
SS
MS
Pseudo-F
P
Perms
9814
Region
1
11875
11875
3.3645
0.0001
2nd Stage Analysis
Habitat
2
14120
7059
1.2151
0.3575
718
Megafaunal community data from slope, canyon, and seamount
habitats from TAN1004 and TAN1206 were compared to those
of seep megafauna and vent megafauna from 27 stations based
on data from previous surveys. Each seep and vent station was
assigned to the nearest depth stratum (700, 1000, or 1200 m)
to allow direct comparisons with slope, canyon, and seamount
sites. Complete matching between sampling depth strata was
not possible for analysis purposes, so instead comparisons had
to made using combined “shallow” (700 and 1000 m) and
“deep” (1200 and 1500 m) depth strata. The effects of habitat
and depth on megafaunal community structure within each
Depth
3
29102
9700
1.5608
0.1078
8806
Frontiers in Marine Science | www.frontiersin.org
Region × Habitat
2
11197
5598
1.5862
0.0001
9761
Region × Depth
3
16295
5431
1.5389
0.0001
9742
Habitat × Depth
6
29290
4881
1.2274
0.1373
9851
Region × Habitat
× Depth
6
22592
3765
1.0668
0.1721
9600
Residual
75
279000
3529
Total
104
452000
Df, degrees of freedom; SS, sum of squares; MS, mean square; Pseudo-F, pseudo Fstatistic; P, probability; Perms, unique permutations; significant results are highlighted in
bold.
6
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
slope or seamount habitats. Taxa in higher abundance in
canyon habitats included Bathysiphon, Brissopsis, Sipuncula,
and M. musculus on the Hikurangi Margin, and Sipuncula,
M. musculus, Spiochaetopterus (polychaete), and Ophiomusium
lymani (ophiuroid) in the Bay of Plenty. On the Hikurangi
Margin, Sympagurus dimorphus (pagurid), Scaphander otagoensis
(gastropod), O. lymani and Benthodytes were more abundant
at slope than seamount habitats, while the opposite was
observed for Fusitriton (gastropod), M. musculus, Ophiozonella
stellamaris (ophiuroid), and Amphioplus. Taxa contributing to
the dissimilarity between slope and seamount habitats in the
Bay of Plenty included Munida (galatheid), Ophiura (ophiuroid),
and Enallopsammia rostrata (scleractinian coral), as well as those
two taxa that were noted above as being responsible for regional
differences in community structure (being more abundant on
seamounts).
TABLE 2 | Results of pairwise PERMANOVA analyses testing for the
effects Habitat (slope, canyon, and seamount) and Depth (700, 1000, 1200,
and 1500 m) on megafauna community structure in Hikurangi Margin and
Bay of Plenty after accounting for the effect of spatial covariates (latitude,
longitude; results not shown).
Hikurangi Margin
t
P
Perms
Bay of Plenty
t
P
Perms
Habitat
slope versus canyon
1.423
0.008
9902
1.225
0.029
9864
slope versus seamount
1.012
0.444
9907
1.521
<0.001
9841
canyon versus seamount
1.387
0.016
9915
1.318
<0.001
9815
Depth
700 versus 1000 m
1.527
0.004
9918
1.394
<0.001
9869
700 versus 1200 m
1.797
0.004
9940
1.624
<0.001
9874
700 versus 1500 m
1.395
0.030
9929
1.871
<0.001
9863
1000 versus 1200 m
1.263
0.045
9902
1.172
0.029
9854
1000 versus 1500 m
1.164
0.124
9908
1.620
<0.001
9849
1200 versus 1500 m
0.866
0.705
9926
1.208
0.034
9870
Environmental Variables Influencing
Community Similarity among Slope,
Canyons, and Seamounts
t, t-statistic; P, probability; Perms, number of unique permutations; significant results are
highlighted in bold.
In slope, canyon and seamount habitats, fishing intensity
was higher on the Hikurangi Margin than in the Bay of
Plenty, and in both regions fishing intensity was highest on
seamounts (Figure 3A). On the Hikurangi Margin, fishing
intensity accounted for most of the environmental dissimilarity
between seamounts and the other non-chemosynthetic habitats
(accounting for 44–58% of environmental dissimilarity between
these habitats, SIMPER), whereas in the Bay of Plenty higher
substrate diversity and occurrence of cobble and gravel substrates
on seamounts accounted for most of the dissimilarity from other
habitats (accounting for 17–27% of environmental dissimilarity)
(Figures 3B–D). Environmental dissimilarity between slope and
canyon habitats on both the Hikurangi Margin and in the Bay
of Plenty was mostly due to the standard deviation of the slope
(both regions, 11–17% of dissimilarity) and rugosity (Hikurangi
Margin only, 11%), both of which were higher at canyon habitat
(Figures 3E,F).
The environmental variables identified as being the most
important in the DistLM marginal tests all featured in the best
step-wise (sequential tests) DistLM models. The best DistLM
model that included both regions explained 30% of the variance
in community structure, and included 13 variables describing
substrate and seafloor topography (Table 3). Depth and substrate
diversity were the most important variables explaining the
difference in megafaunal community structure across the two
regions. The best DistLM model for habitats on the Hikurangi
Margin explained 37% of the variance and included five variables,
with boulders, bedrock, and depth being the most important.
The variables with the highest correlation with community
structure at habitats in the Bay of Plenty were depth, substrate
diversity and shell hash, which together with three more variables
explained 23% of the variance observed in this region. Trawl
intensity did not feature in any of the best DistLM models for
both or individual regions, despite being significantly correlated
with community structure in marginal tests for the Bay of
Plenty.
between the two study regions (Figure 2A), but a less distinct
separation of samples by habitat and depth within each region
(Figures 2B–E).
SIMPER analysis indicated that dissimilarity between
communities in the two study regions was accounted for
by small contributions (<2%) by a large number of taxa
(Supplementary Table 4). At slope habitats, most of the taxa
contributing to dissimilarity between the regions (96%) were
more abundant, or were only found at, the Hikurangi Margin.
These taxa included Plutonaster knoxi, Zoroaster alternicanthus
(asteroids), Gracilechinus multidentatus, Hygrosoma luculentum
(echinoids), and Benthodytes (holothuroid). At canyon habitats,
taxa characterizing the dissimilarity between regions (93%) were
also more abundant or only present at the Hikurangi Margin.
These taxa included a diverse range of taxonomic groups e.g.,
Brissopsis (echninoid), Sipuncula (sipunculid), Bathysiphon
(foraminiferan), Amphioplus (ophiuroid), Molpadia musculus
(holothuroid), and Eunice (polychaete). For seamount habitat,
while most of the taxa contributing to the dissimilarity between
regions (96%) were also more abundant or only present at
Hikurangi Margin sites (e.g., the asteroids Z. alternicanthus and
P. knoxi), there were also some taxa that were notably more
abundant or only present at seamounts in the Bay of Plenty.
These taxa include Nematocarcinus (caridean shrimp) and
Nassarius ephamillus (gastropod).
The SIMPER analysis indicated levels of dissimilarity of 86%
(slope v canyon), 82% (canyon versus seamount), and 80%
(slope versus seamount) on the Hikurangi Margin, and 92, 95,
and 96% for the same comparisons in the Bay of Plenty. In
all cases, community dissimilarity was accounted for by small
contributions of dissimilarity (<2.5%) by a large number of
taxa (Supplementary Tables 5, 6). Most of the dissimilarity
between habitats resulted from a number of taxa being more
abundant or present only at canyon habitats compared to
Frontiers in Marine Science | www.frontiersin.org
7
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
FIGURE 2 | MDS ordination plots showing megafaunal community similarity among samples coded for (A) region, (B) habitat, and (C) depth for the
Hikurangi Margin region, and (D) habitat and (E) depth for the Bay of Plenty region.
Community Similarity Including
Chemosynthetic Habitats
distinct from one another than those on the Hikurangi Margin,
with vent communities being the most dissimilar from those in
other habitats.
Community dissimilarity between vents and other deep-sea
habitats in the Bay of Plenty at both shallow and deep strata
was very high (SIMPER, 97–100%), and the taxa contributing
most to this dissimilarity were those endemic to vent habitats:
Alvinocaris niwa (shrimp) and Gigantidas gladius (mussel), and
Alvincocaris longirostris and Vulcanolepas osheai (barnacle), in
the shallow and deep strata, respectively (Supplementary Table 7).
Levels of dissimilarity between communities at seeps and the
other habitats on the Hikurangi Margin (shallow strata only)
were less but still high (89–93%), with the distinguishing taxa
being those that mostly occurred in relatively low abundance
at seep habitat compared to the other habitats on the margin
The second-stage PERMANOVA analysis showed a Habitat effect
in the Bay of Plenty region, with significant differences for
all pairwise habitat comparisons except slope versus canyon
(Table 4). On the Hikurangi Margin, community structure was
significantly different between seep and the other three habitats,
and slope and canyon habitats at 700–1000 m depth while at
1200–1500 m depth, where seeps were not present, only the
comparison between slope and canyon habitats was significant
(Table 4). MDS plots illustrate the patterns underlying the
PERMANOVA analysis, showing the dissimilarity of samples
from seep and vent habitats compared to those from other
habitats within regions (Figures 4A,B). The plot shows that
communities at habitats in the Bay of Plenty region are more
Frontiers in Marine Science | www.frontiersin.org
8
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
FIGURE 3 | Fishing intensity and physical characteristics of canyon, slope, and seamount habitats at the Hikurangi Margin (shaded bars) and Bay of
Plenty (open bars) regions. (A) Fishing intensity, (B) Substrate diversity, (C) Cobble, (D) Gravel, (E) Standard deviation of slope, (F) Seafloor rugosity. Error bars are
standard errors.
were distinct through having relatively high proportions
of deposit-, suspension-, and filter-feeders (Supplementary
Table 9).
(e.g., S. dimorphus), rather than the presence of seep endemic
megafauna (Supplementary Table 8).
Trophic structure did not differ between habitats in the
shallow strata on the Hikurangi Margin but was significantly
different between canyons and the other topographic habitats
in the deep strata. In the Bay of Plenty, trophic structure
was significantly different among all habitats in the shallow
strata, and between vents and all other habitats in the
deep strata (PERMANOVA, Table 5). At the shallow strata
in the Bay of Plenty, seamount habitats had the greatest
proportion of suspension- and filter-feeding taxa. Canyon
habitats also had a relatively high proportion of these
two feeding types but communities here also contained a
high proportion of deposit-feeders and predator/scavengers,
and even some chemoautotrophic taxa. Slope habitats were
dominated by deposit-feeders and predator/scavengers but also
with a notable proportion of suspension-feeders, while vent
habitats were, not surprisingly, dominated by chemoautotrophic
fauna (Figure 5). On the Hikurangi Margin, canyon habitats
Frontiers in Marine Science | www.frontiersin.org
DISCUSSION
The principal aims of this study were to determine levels of
dissimilarity among megafaunal communities in different deepsea habitats, and to assess the relative vulnerabilities of these
communities to anthropogenic disturbance. Previous studies of
deep-sea megafauna among habitats have generally focused on
particular taxa (e.g., only decapods) or have included fish as well
as invertebrate megafauna, and their study habitats have generally
been located at different depths. Analysis methods also differ. For
example, initial analysis of faunal communities has sometimes
been used to define habitats, rather than using a priori physical
habitat identification, and different taxonomic resolution and
similarity metrics have been used. This limits direct comparisons
between this study and these earlier findings, but nonetheless we
9
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
Margin than in the Bay of Plenty. Differences in food availability
between the two regions, indicated by higher levels of surface
water chorophyll and sediment chlorophyll/phaeopigment
concentration on the Hikurangi Margin (Leduc et al., 2016; Rosli
et al., 2016), are likely to be mainly responsible for this observed
pattern. Food availability is a major driver of structure in
deep-sea benthic communities at regional scales, where elevated
levels (up to a point) increase the overall faunal abundance
and relative abundance of particular taxa (Levin et al., 2001;
Smith et al., 2008). Rowden et al. (2010b) also attributed the
regional differences they observed in squat lobster communities
to the relative levels of particulate organic matter flux to the
seafloor (another proxy for food availability). In the present
study, substrate diversity was also markedly different between
regions, being greater at canyons and seamounts in the Bay of
Plenty, and at slopes in the Hikurangi Margin region. Substrate
diversity has been used to explain megafaunal community
structure in previous studies; heterogeneous substrates can
provide a larger number of niches to be occupied by fauna
and thereby influence the community composition (e.g., on
seamounts, Raymore, 1982). Usually, community structure is
related to habitat heterogeneity at relatively small spatial scales,
and this is confirmed by the finding that substrate diversity (after
depth) was the most important variable explaining differences in
community structure among all samples. However it is possible
that the regional-scale differences in substrate diversity could
partly explain the levels of community dissimilarity observed
between the study regions.
Other factors could also be responsible for the difference in
community structure across habitats in the two regions. Fishing
intensity was consistently and markedly higher (12–114 times)
on the Hikurangi Margin than in the Bay of Plenty. While fishing
was measured simply as the total number of trawls at a site in
a 15-year period before sampling, together with what is known
about the history of trawling for the region (authors unpublished
data), it is an indication that many places on the Hikurangi
Margin are probably chronically disturbed by bottom trawling.
Benthic communities at trawled habitats are often characterized
by relatively few megafaunal taxa but with high abundances of
some opportunistic species that are able to survive and flourish
under chronic disturbance (Ramsay et al., 1996; Collie et al.,
1997). Thus fishing disturbance could partly explain the generally
higher abundances observed at slope, canyon and seamount
habitats on the Hikurangi Margin compared to the Bay of Plenty
(Supplementary Table 4).
TABLE 3 | Results of distance-based linear model (DistLM) analyses
showing correlations between predictor variables and megafaunal
community structure (sequential tests) for Hikurangi Margin and Bay of
Plenty regions combined, and each region separately.
SS
Pseudo-F
P
R2
R2 Cumul.
res.df
BOTH REGIONS
+ Depth
18543
4.744
0.001
0.04
0.13
101
+ Substrate diversity
15332
4.0408
0.001
0.03
0.16
100
+ Profile curvature
7470.7
1.9884
0.001
0.02
0.18
99
+ Shell hash
6330.7
1.6968
0.001
0.01
0.19
98
+ Boulders
5748.4
1.5494
0.002
0.01
0.20
97
+ Cobbles
6711.3
1.8243
0.001
0.01
0.22
96
+ Rugosity
5709.8
1.5611
0.002
0.01
0.23
95
+ SD depth
5299.8
1.456
0.007
0.01
0.24
94
+ Bedrock
4825
1.3302
0.035
0.01
0.25
93
4915.6
1.3604
0.03
0.01
0.26
92
+ Gravel
+ SD slope
5091
1.4153
0.009
0.01
0.28
91
+ Sand
4927.2
1.3754
0.023
0.01
0.29
90
+ Depth range
4561.4
1.2773
0.045
0.01
0.30
89
HIKURANGI MARGIN
+ Boulders
9103.2
2.9757
0.001
0.08
0.19
32
+ Bedrock
6417.8
2.1749
0.001
0.05
0.24
31
+ Depth
6355.8
2.2401
0.001
0.05
0.29
30
+ Sand
4718.3
1.7018
0.004
0.04
0.33
29
+ SD slope
4060.6
1.4893
0.033
0.03
0.37
28
+ Depth
15624
3.8429
0.001
0.05
0.12
65
+ Substrate diversity
8028.1
2.0051
0.001
0.03
0.14
64
+ Shell hash
7883.7
1.9998
0.001
0.03
0.17
63
+ Depth range
6294.9
1.6123
0.002
0.02
0.19
62
+ SD slope
5610.9
1.4474
0.011
0.02
0.21
61
+ Plan curvature
5198.3
1.3487
0.028
0.02
0.23
60
BAY OF PLENTY
Spatial covariates (latitude and longitude) were entered first before fitting other variables.
[SS, sum of squares; Pseudo-F, pseudo F-statistic; P, probability; R2 , proportion of
explained variation attributable to each variable; R2 Cumul., cumulative proportion
of variation; res. Df, residual degrees of freedom; SD, standard deviation; significant
contributions to the models are highlighted in bold].
have considered the latter in the discussion below whenever they
are relevant.
Differences in Communities between
Regions
The most obvious finding of the present study was that
megafaunal communities of the two study regions are
significantly different. This difference between regions was
greater than those among habitats or depth strata, with many
taxa contributing across all topographic habitats. Rowden et al.
(2010b) observed a regional difference in deep-sea squat lobster
communities (sampled by sled, dredge, and trawl) that was
also greater than dissimilarity among habitats, with levels of
community dissimilarity among regions between 87–100%
(Rowden unpublished data.).
Generally, fauna characterizing the habitat difference between
the two study regions were more abundant on the Hikurangi
Frontiers in Marine Science | www.frontiersin.org
Differences in Communities among Slope,
Canyon, and Seamount Habitats
(Hypothesis 1)
Megafaunal community structure on the Hikurangi Margin was
significantly different only between canyon and both slope and
seamounts habitats, whereas in the Bay of Plenty it differed
among all three habitats. These differences among habitats
were generally consistent across all depth strata, and levels
of community dissimilarity among habitats were generally
high: 80–86 % on the Hikurangi Margin and 92–96% in the
10
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
TABLE 4 | Results of pairwise PERMANOVA analyses testing for the effects Habitat (slope, canyon, seamount, and seep or vent habitats) on community
structure in Hikurangi Margin and Bay of Plenty regions.
Depth
Hikurangi Margin
t
700 and 1000 m
1200 and 1500 m
Bay of Plenty
P
Perms
t
P
Perms
Slope, canyon
1.2335
0.026
1708
1.1774
0.0605
5658
Slope, seamount
0.89924
0.6935
1706
1.7602
0.0001
9661
Slope, seep/vent
1.6296
0.0013
1706
2.0319
0.0005
2902
Canyon, seamount
1.1516
0.1155
1710
1.7698
0.0001
9849
Canyon, seep/vent
1.7771
0.0006
1707
2.0997
0.0001
8857
Seamount, seep/vent
1.7574
0.0015
1709
2.3259
0.0001
9856
Slope, canyon
1.3962
0.0179
462
1.116
0.1632
8495
Slope, seamount
0.97219
0.5665
208
1.3949
0.002
9574
Slope, seep/vent
Canyon, seamount
–
–
–
2.1289
0.0001
8977
1.312
0.0668
210
1.2469
0.0091
9837
Canyon, seep/vent
–
–
–
2.1429
0.0001
9888
Seamount, seep/vent
–
–
–
2.1396
0.0001
9894
Samples from the two shallow (700 and 1000 m depth), and deep strata (1200 and 1500 m) were combined to provide sufficient replication for analyses. No samples were obtained
from seep habitat in the deep strata of Hikurangi Margin. [t, t-statistic; P, probability; perms, number of unique permutations; significant results indicated in bold].
megafaunal decapod crustaceans sampled by trawl at open
slope and canyon habitats on the northwestern margin of the
Mediterranean Sea (off Spain), and found that community
dissimilarity between canyon habitat (430–515 m) and the slope
habitats (∼250–700 m) was 43–60% in autumn and 64–85%
in spring. In a study that compared squat lobster communities
from different deep-sea habitats in the southwestern Pacific
Ocean, Rowden et al. (2010b) observed significant differences
between seamount and rise and ridge habitat (in some regions),
but not between seamounts and slope and plateau habitats at
comparable depths. Levels of dissimilarity among significantly
different habitat communities ranged from 53 to 92% (Rowden
unpublished data). Ramirez-Llodra et al. (2010) compared
community similarity of megafauna (including fish) sampled
by trawl among two canyon sub-habitats (head and wall) and
open slope habitat (samples from different times of the year,
but analyzed together). Community structure was significantly
different between the three habitats (Ramirez-Llodra et al., 2010)
but with levels of dissimilarity among habitats ranging between
40 and 48% (Ramirez-Llodra pers comm.). Kelly et al. (2010)
compared community similarity of deep-sea habitats in the
Gulf of Maine (northeast USA/Canada), using presence-absence
data that included megafauna sampled by trawls. Comparisons
among habitats were not controlled for depth, so dissimilarity
values might be expected to be higher than those observed in
the present study. Generally, habitats that were further apart
had highly dissimilar fauna (>90%: shelf edge and channel
versus rise and seamount), whereas those habitats that were
physically closer had a less dissimilar fauna (62–75%: shelf versus
channel, canyon versus slope, seamount versus rise). Overall,
these comparisons indicate that megafaunal communities of
deep-sea habitats can vary considerably, and where levels of
dissimilarity are high, habitats may not be well connected.
However, high dissimilarity among habitats is not always
found.
FIGURE 4 | MDS ordination plots showing megafaunal community
similarity among samples at habitats on (A) the Hikurangi Margin, and in
(B) the Bay of Plenty.
Bay of Plenty. Although not directly comparable, levels of
community dissimilarity observed among habitats in previous
studies have generally been lower. Cartes et al. (1994) examined
Frontiers in Marine Science | www.frontiersin.org
11
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
TABLE 5 | Results of pairwise PERMANOVA analyses testing for the effects Habitat (slope, canyon, seamount, and seep or vent habitats) on trophic
structure in Hikurangi Margin and Bay of Plenty regions.
Depth
700 and 1000 m
1200 and 1500 m
Hikurangi Margin
Bay of Plenty
t
P
Perms
t
P
Perms
Slope, canyon
1.1286
0.2507
1709
1.8161
0.011
5755
Slope, seamount
0.85954
0.5856
1710
3.1218
0.001
9704
Slope, seep/vent
1.2689
0.1548
1711
2.1105
0.004
2896
Canyon, seamount
1.102
0.2784
1712
1.5399
0.041
9931
Canyon, seep/vent
1.5553
0.066
1705
1.9672
0.004
8915
Seamount, seep/vent
1.0595
0.321
1704
3.2895
0.001
9915
8498
Slope, canyon
1.7474
0.0223
462
1.3026
0.1243
Slope, seamount
0.89275
0.4649
210
1.4273
0.0822
9671
Slope, seep/vent
–
–
–
2.6441
0.0002
9034
Canyon, seamount
1.9922
0.0395
210
0.96245
0.4541
9935
Canyon, seep/vent
–
–
–
2.6777
0.0001
9899
Seamount, seep/vent
–
–
–
3.1512
0.0001
9945
Samples from the two shallow (700 and 1000 m depth), and deep strata (1200 and 1500 m) were combined to provide sufficient replication for analyses. No samples were obtained
from seep habitat in the deep strata of Hikurangi Margin. [t, t-statistic; P, probability; perms, number of unique permutations; significant results indicated in bold].
Canyon habitats were distinguished from slope and seamount
habitats in both regions by a range of taxa that were either present
in higher abundance in canyons, or were absent from other
habitats. Some of these more abundant taxa differed between the
two regions, reflecting the regional differences discussed above
but others were the same in both regions and, thus, likely to be
more revealing of the reason for the differences among canyons
and other habitats. Sipuncula and M. musculus are both infaunal
deposit feeders and their higher abundances in canyon habitats
probably reflect the elevated levels of organic matter that can
occur in canyons, and which have been observed previously to
support high densities of these types of deposit-feeding organism
in a New Zealand canyon (De Leo et al., 2010). Differences in
community structure between canyon habitats and other habitats
in both regions could also be related to the higher topographic
complexity of the seafloor at canyons (indicated by higher values
of the SD of the slope, and rugosity), as suggested by Schlacher
et al. (2007) as a reason for the distinct nature of megafaunal
communities in canyons.
Soft sediments are generally more prevalent on continental
slopes than on seamounts, and seamounts in the Bay of Plenty
had much higher coverage of hard substrates such as cobbles
and gravel, and higher substrate diversity. The latter, together
with shell hash, another category of hard substrate, were two of
the environmental variables that explained the most variability
observed in community structure in the Bay of Plenty. Taxa
that contributed to differences between slope and seamount
communities in this region reflect the differences in substrate
type. M. musculus and O. lymani are associated with muddy
or sandy soft sediments (Grassle et al., 1975) and were more
abundant in slope habitats. In contrast the taxa that were more
abundant at seamount than slope habitat in the Bay of Plenty,
and contributed to the observed dissimilarity, included those
that are more often associated with hard or mixed substrates
(E. rostrata, Munida, N. ephamillus, Ophiura, Nematocarcinus
Frontiers in Marine Science | www.frontiersin.org
(see similar taxa in Howell et al., 2010a). The lack of a difference
in community structure between slope and seamount habitats on
the Hikurangi Margin could reflect the sedimentary nature of
seamounts on that margin, and their structural similarity to slope
habitat in that region (Bowden et al., 2016).
Differences in Communities Including
Chemosynthetic Habitats (Hypothesis 1)
It is not surprising that megafaunal communities at vent and
seep habitats were found to be different from those at other
deep-sea habitats. However, the extent of this difference between
chemosynthetic-based communities and those of surrounding
deep-sea habitats is rarely determined in a comparable manner,
and levels of dissimilarity are rarely reported. In the present
study, levels of dissimilarity between vent and seep communities
and those at other deep-sea habitats ranged between 97–100%
and 89–93%, respectively. This difference between the two
chemosynthetic-based habitats probably reflects differences in
both environmental characteristics of vents and seeps and their
associated chemoautotrophic fauna. Benthic communities at the
more heterogeneous seep habitats have been shown previously
to consist of a greater proportion of “background” fauna than
those communities associated with the more discrete vent
habitats (Bernardino et al., 2012) and the main megafaunal
chemoautotrophic species at seeps on the Hikurangi Margin
are either wholly or partially infaunal (Baco et al., 2010), and
thus have lower catchabilities in sleds and trawls than the
predominantly epifaunal taxa associated with vents. The present
analysis showed that differences in communities between seep
and other habitats were largely explained by lower abundances
of taxa that are more common at the other habitats, rather
than presence of seep-endemic taxa. Similarly, the present
analysis supported the expectation that the dissimilarity between
vent communities and those of other deep-sea habitats was
mostly determined by the presence of vent-endemic taxa. The
12
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
FIGURE 5 | Mean proportions of feeding groups for megafauna at seamount, canyon, slope and vent habitats in the Bay of Plenty region at the
shallow strata (700 and 1000 m). Trophic structure was not significantly different between habitats in the Hikuangi Margin at the shallow strata, so corresponding
results are not shown.
dissimilarity between communities at seep and vent habitats
themselves was almost 100%, which is comparable to the highest
levels of dissimilarity observed between megafaunal communities
at seeps and vents in the northwest Pacific (off Japan). Nakajima
et al. (2015) identified three vent and five seep communities in
this region, with community dissimilarity levels between the two
types of habitat ranging from 85 to 100%, being highest when
the two habitats were relatively distant (geographically and/or
bathymetrically) (Nakajima pers. comm.). The vent and seep
habitats compared in the present study are also geographically
distant, and seep sites were not as deep as some of the vent sites.
Not surprisingly, the analysis for differences in trophic
structure among all habitats demonstrated the dominance of
the chemoautotrophic feeding group at vent and seep habitats,
but this group was relatively more dominant at vent than
at seep habitats (Supplementary Table 9). This observation is
partly related to the physical characteristics of the two habitats
(see above), but also by the differences in the chemicals and
fluid regime at seeps and vents. Benthic communities at seep
habitats have been shown to consist of heterotrophic or partly
heterotrophic fauna, as well as the specialized chemoautotrophic
fauna (Levin et al., 2003), whereas, heterotrophic fauna are less
common in the more toxic vent habitats (Bernardino et al.,
2012).
Frontiers in Marine Science | www.frontiersin.org
Community structure in canyons was also significantly
different from that on the slope at both shallow and deep strata
on the Hikurangi Margin. As noted in the previous section, this
is probably the result of the relatively higher food availability,
more complex topography, and availability of hard substrates
in canyons compared with slopes. Trophic structure differed
between these two habitats, but only at the deep strata, where
canyon habitats had a higher proportion of deposit-, suspension, and filter-feeding taxa than slope habitats, while the latter had a
larger proportion of grazers and scavengers/predators. As already
noted, deposit feeders are likely to be particularly common
in canyons where organic matter accumulates, while sessile
megafauna that attach to the hard substrates found on the canyon
walls can occur in relatively high abundance benefitting from
increased food availability and delivery resulting from canyonassociated hydrodynamic features such as upwelling (Duineveld
et al., 2001). The dominance of the other feeding groups at
slope habitats could relate to the homogenous nature of the
slope substrate providing a more extensive habitat over which the
typically mobile grazers, scavengers and predators can search to
find food.
In the Bay of Plenty, however, this difference between
communities in canyon and slope habitats was not seen, while
there was a significant difference between slope and seamount
13
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
Relative Vulnerability of Communities at
Different Deep-Sea Habitats (Hypotheses 2
and 3)
communities at both depth strata. The latter finding is probably
the result of the contrasting substrate conditions between
slope and seamounts in this region, and analysis of the trophic
structure of the megafaunal communities supports this finding,
indicating that seamount habitats had the greatest proportion of
suspension- and filter-feeding taxa, and although suspensionfeeders were also present on slopes, communities at this habitat
were dominated by deposit-feeders and predator/scavengers.
Seamounts are typically dominated by hard substrates that
provide habitat for sessile suspension and filter-feeders, such
as corals and sponges (Samadi et al., 2007). Benthic fauna
in these trophic groups can occur in high abundance on
seamounts, particularly where the morphology of seamounts
influences hydrodynamic conditions that enhance food delivery
to suspension and filter-feeders (Genin et al., 1986). The
seafloor at slope habitats, as already noted, is predominantly
composed of soft sediment, which is why deposit-feeders
dominate the megafauna found at this habitat in the Bay of
Plenty. The relatively high proportion of predators/scavengers
in slope habitats could relate to the homogenous slope
substrate providing a more favorable habitat for such
mobile taxa.
The distinct nature of communities in vent and seep habitats,
makes them potentially more vulnerable to disturbance than
those at slopes, canyons and seamounts. Vent-endemic taxa
can only recover from human disturbance through larval
recruitment from other vent habitats. Vent communities can
recover relatively rapidly from catastrophic natural disturbances
(e.g., Tunnicliffe et al., 1997) because the larvae of many ventendemic taxa disperse widely (e.g., Marsh et al., 2001; Levin et al.,
2016). However, recovery can also be slow where populations of
chemoautrophic vent fauna depend on local source populations
for their long-term maintenance in a region (Metaxas, 2011).
For example, populations of the vent mussel G. gladius (one
of the characterizing species of communities at vent habitats
in the present study) appear to depend on a source population
on Rumble V seamount (Boschen et al., 2015), which means
that individual vent communities may be more vulnerable to
disturbance than simple recovery models for vent communities
predict. Communities at seep habitats were not as dissimilar
as those at vents were from the other deep-sea habitats, thus
they may be less vulnerable to disturbance. The characterizing
taxa of seep habitats included taxa that were found in adjacent
habitats, and hence the other habitats (assuming they are not also
disturbed) can act as a colonizing source of such fauna. However,
seep communities also possess chemoautotrophic fauna, which
like vent-endemic fauna, depend on larval recruitment from
other seep habitats for recovery following disturbance. Some
characteristic seep taxa are only found at seep habitats, while
others are also found at other reducing environments such as
highly reduced organic muds, and wood, kelp, and whale falls
(e.g., Bernardino et al., 2012). This means that the pool of
potential colonizers for a disturbed seep habitat may be greater
than for a disturbed vent habitat.
Seep and vent habitats are also relatively small in extent and
thus the more localized communities may be more susceptible
to impacts from human activities than the larger slope, canyon,
and seamount habitats. That is, the scale of the disturbance will
likely be more concentrated, and the fauna of patch or islandlike habitats are particularly prone to extinction from disturbance
impacts because of their relatively small population sizes and the
source of colonizers for a population to recover will almost all
have to come from a distant and non-contiguous source habitat
(if the site is not part of a local “field” of such habitats) (Hanski,
1999). In the New Zealand context, the benthic communities of
seep and vent habitats will also be more vulnerable to disturbance
because they occur within a relatively small biogeographic
province (Baco et al., 2010; Rogers et al., 2012) compared to
the other deep-sea habitats which occur within a much broader
biogeographic province (Watling et al., 2013). Thus, the species
pool of potential colonizing fauna following disturbance is much
larger for slope, canyon and seamount communities than for
those of seeps and vents.
Levels of community dissimilarity were also high when just
comparing among megafauna at canyon, slope and seamount
Relative Vulnerability of Communities in
Different Regions
Differences between the two study regions were driven largely
by changes in abundances of many taxa, indicating the two
regions share overlapping species pools. An implication of this
is that they might have similar vulnerabilities to disturbance.
However, the more abundant populations on the Hikurangi
Margin, where food availability is higher, are likely to be less
vulnerable than less abundant populations in the Bay of Plenty.
Vulnerability will vary with the ecological characteristics of
different taxa (e.g., Tyler-Walters et al., 2009), but assuming a
constant mortality rate following the onset of a disturbance, and
similar distributional extent, less abundant populations are more
likely to become locally extinct than more abundant populations,
thus requiring more immigration from undisturbed areas to
return to pre-disturbance levels. Recovery from disturbance
depends largely on availability of colonizers, which consist
mainly of motile adults from adjacent undisturbed areas during
the initial recovery phase, but also planktonic larvae during
later recovery (Zajac and Whitlach, 1991), which in turn
is influenced by population size. Elevated food availability
also increases population growth (Sibly and Hone, 2002),
thereby accelerating recovery of benthic communities following
disturbance. Therefore undisturbed megafaunal communities in
the Bay of Plenty may be more vulnerable to human impact
than Hikurangi Margin communities despite being characterized
by the same taxa. However, as noted above, it is possible that
the higher abundances of some fauna on the Hikurangi Margin
could be as a result of chronic disturbance from bottom trawling.
Chronically disturbed communities may be more vulnerable to
disturbance if they are close to a threshold/tipping point that
precipitates an alternate community state (Hewitt and Thrush,
2010).
Frontiers in Marine Science | www.frontiersin.org
14
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
communities in regions that have already been chronically
disturbed may also require protection (especially if believed close
to an ecological tipping point), in general there is a greater need
to afford protection on a regional basis to relatively un-impacted
benthic communities—e.g., in the Bay of Plenty. Such protection
in a region is particularly important where future disturbance
could occur, or where the nature and extent of a new source of
human disturbance could be different. The threat of disturbance
from future seabed mining or drilling is potentially similar in
the two regions (see Introduction, and below), and this potential
should also be considered in the design of spatial management
measures.
The evaluation of faunal vulnerability here is not a formal
risk assessment, which would involve consideration of a more
detailed range of sensitivity and recoverability traits (e.g., Hewitt
et al., 2011). Nevertheless, megafaunal communities of vent and
seep habitats were assessed to be more vulnerable to disturbance
than those of slope, canyon and seamount habitat because of
the potential limited ability of chemoautrophic taxa to recolonise
areas after disturbance, the more localized nature of seep and
vent habitats, and their smaller-scale biogeography. These factors
suggest that chemosynthetic-based communities may require a
different form of spatial management to that which could be
employed to protect some other forms of deep-sea habitats. This
distinction has already been recognized, and specific guidelines
(the “Dinard Guidelines”) exist to help design spatial protection
measures for these ecosystems (Van Dover et al., 2012).
The majority of vent habitats in New Zealand waters are
already included in Benthic Protection Areas (BPAs) (Helson
et al., 2010), or Seamount Closure Areas (Brodie and Clark, 2003)
that prohibit the use of bottom trawling, and some of these sites
may soon receive further protection through inclusion in the
proposed Kermadec Ocean Sanctuary (http://www.mfe.govt.nz/
marine/kermadec-ocean-sanctuary). However, these protected
sites have not been selected according to the “Dinard Guidelines,”
and some vent habitats in the Bay of Plenty region remain
exposed to the impact of any future seabed mining for SMS
deposits. None of the more than 30 seep habitats that have
been found on the Hikurangi Margin (Greinert et al., 2010)
are protected by BPAs, and an impact on benthic community
structure from trawling has already been observed (Baco
et al., 2010; Bowden et al., 2013), which could potentially be
compounded by future drilling for gas hydrate deposits (Bowden
et al., 2013). Thus, in a New Zealand context, seep habitats on
the Hikurangi Margin are in particular need of consideration
for protection, and this should be undertaken according to the
guidelines designed specifically for chemosynthetic ecosystems.
The assessment of communities from slopes, canyons and
seamounts concluded that topographic habitat type by itself
might not be a particularly useful basis on which to determine
relative vulnerability to anthropogenic disturbance, and hence
to design spatial management measures. Designing a network of
MPAs according to classifications of topographic habitat type has
been advocated (e.g., for deep-sea habitats off Australia, Harris,
2007), but the lack of a reliable distinction between megafaunal
communities of such habitats raises questions about the general
utility of this approach (this study, O’Hara et al., 2008; Williams
habitats, but the relative values were not the same between pairs
of habitats in each study region. This finding means that relative
vulnerability of deep-sea habitats to anthropogenic disturbance
cannot be consistently defined by habitat type using levels
of observed dissimilarity. Furthermore, community structure
and trophic structure were not always significantly different,
which further suggests that the vulnerabilities of megafaunal
communities of some deep-sea habitats are likely to be similar
(for example, slope and seamount communities on the Hikurangi
Margin). However, community structure at canyon habitats was
always significantly different from those at slope and seamount
habitats, and trophic structure was also significantly different at
this habitat at shallow depths in the Bay of Plenty (when also
comparing vent and seep habitats) and at greater depths on the
Hikurangi Margin. Canyon communities were characterized by
an abundant megafauna, many taxa of which were also found at
other habitats in lower abundances. Thus, canyon communities
may have the ability to recover more quickly from disturbance
than slope and seamount communities. Furthermore the higher
complexity of the seafloor topography observed at canyon
habitats could make bottom trawling difficult and provide refuges
for megafauna from such disturbance. Seamount habitats can
also have a diverse substratum that could provide similar refuges
from disturbance (Williams et al., 2010b), and the communities
of canyons, like the seamounts, had a high proportion of
suspension- and filter-feeding taxa (e.g., stony corals), which
previous studies have shown to be vulnerable to disturbance by
bottom trawling (Althaus et al., 2009; Clark and Rowden, 2009).
Thus, the results of the present study demonstrate that it is not
straightforward to assess the relative vulnerability of megafaunal
communities at canyons, slopes and seamounts on a habitat basis
alone.
Implications for the Management of
Anthropogenic Disturbance to Deep-Sea
Habitats
If deep-sea habitats in different regions (separated in the case of
the present study by about 500 km) are not equally vulnerable
to disturbance from human activities, it has implications for the
design of spatial management measures to protect biodiversity.
Design guidelines for the distribution of marine protected areas
(MPAs) generally recommend a regional approach, but this is
based on the need to represent biogeographic differences in
benthic fauna (e.g., Roberts et al., 2003). The results of the present
study suggest that such guidelines should consider more than
just differences in the species pool of an area. For example,
the megafaunal communities of deep-sea habitats in the Bay of
Plenty, despite sharing most species with the Hikurangi Margin,
are considered by our assessment to be more vulnerable to
disturbance than those on the Hikurangi Margin—an assessment
affected by patterns in abundance and feeding mode. The two
regions are already subject to different impacts from fishing—
with higher bottom trawl intensity on the Hikurangi Margin
than in the Bay of Plenty. Thus it is probable that megafaunal
communities at deep-sea habitats in the former region are already
impacted (Cryer et al., 2002; Bowden et al., 2013). Although
Frontiers in Marine Science | www.frontiersin.org
15
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
ministry. We thank the officers, crew and scientific personnel
of RV Tangaroa voyages TAN1004 and TAN1206, the principal
sampling voyages of this project, as well as those that took part
in voyages that were the source of additional data—TAN0107;
TAN0616; TAN1104. Funding for these latter voyages was
provided by the former New Zealand Foundation for Research,
Science and Technology (CO1X0028), NOAA Ocean Exploration
(NA17RJ1231 and NA050AR417076), and Land Information
New Zealand (also CO1X0906), respectively. Thanks to Owen
Anderson and Arne Pallentin (both NIWA) for processing the
trawl effort data and multibeam data, respectively. We are
also grateful for the unpublished data on levels of community
similarity supplied by Eva Ramirez-Llodra (NIVA, Norway) and
Ryota Nakajima (formally of JAMSTEC, Japan). Particular thanks
to Kareen Schnabel and Sadie Mills of the NIWA Invertebrate
Collection who organized the curation and distribution of
samples for identification. Lastly we acknowledge the crucial
contribution of the many taxonomists and parataxonomists who
identified the megafauna (see Supplementary Table 2 for their
names and affiliations).
et al., 2009; Howell et al., 2010b). The identification of habitat
type is, nevertheless, a useful first stage in the spatial management
process (for example, it can indicate where sensitive fauna are
more likely to occur), but more detailed site-specific assessments
of the composition, abundance, and relative vulnerability of
communities at deep-sea habitats are required to design the
most effective spatial management measures within any region
(Howell et al., 2010a).
AUTHOR CONTRIBUTIONS
AR contributed to the design of the study, data analysis, and
report writing. DL contributed to the design of the study, data
collection, data analysis, and report writing. MC contributed to
the design of the study, data collection, and report writing. DB
contributed to the design of the study, data collection, and report
writing.
ACKNOWLEDGMENTS
This study is part of NIWA research project Impact of resource
use on vulnerable deep-sea communities funded by the New
Zealand Ministry for Business, Innovation and Employment
(CO1X0906). The study also represents part of NIWA’s
contribution to GNS research project Harnessing New Zealand’s
Gas Hydrate Resources (CO5X1204), funded by the same
SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found
online at: http://journal.frontiersin.org/article/10.3389/fmars.
2016.00241/full#supplementary-material
REFERENCES
to human activities. PLoS ONE 8:e76869. doi: 10.1371/journal.pone.00
76869
Brodie, S., and Clark, M. (2003). “The New Zealand Seamount Management
Strategy – steps towards conserving offshore marine habitat,” in Aquatic
Protected Areas: What Works Best and How Do We Know? Proceedings of the
World Congress on Aquatic Protected Areas, Cairns, Australia, August 2002, eds
J. P. Beumer, A. Grant, and D. C. Smith (Cairns: Australian Society of Fish
Biology), 664–673.
Cartes, J. E., Company, J. B., and Maynou, F. (1994). Deep-water decapod
crustacean communities in the Northwestern Mediterranean: influence
of submarine canyons and season. Mar. Biol. 120, 221–229. doi:
10.1007/BF00349682
Clark, M., and O’Driscoll, R. L. (2003). Deepwater fisheries and aspects of their
impact on seamount habitat in New Zealand. J. Northwest Atl. Fish. Sci. 31,
441–458.
Clark, M. R. (1995). “Experience with the management of orange roughy
(Hoplostethus atlanticus) in New Zealand, and the effects of commercial fishing
on stocks over the period.1980–1993,” in Deep-Water Fisheries of the North
Atlantic Oceanic Slope, ed A. G. Hooper (Alphen aan den Rijn: Kluwer),
251–266.
Clark, M. R., and Rowden, A. A. (2009). Effect of deepwater trawling on
the macro-invertebrate assemblages of seamounts on the Chatham Rise,
New Zealand. Deep Sea Res. I 56, 1540–1554. doi: 10.1016/j.dsr.2009.
04.015
Clark, M. R., and Stewart, R. (2016). The NIWA seamount sled: an effective
epibenthic sledge for sampling epifauna on seamounts and rough seafloor. Deep
Sea Res. I 108, 32–38. doi: 10.1016/j.dsr.2015.12.005
Clarke, K. R., Chapman, M. G., Somerfield, P. J., and Needham, H. R.
(2006a). Dispersion-based weighting of species counts in assemblage
analyses. Mar. Ecol. Prog. Ser. 320, 11–27. doi: 10.3354/meps
320011
Clarke, K. R., and Gorley, R. N. (2006). PRIMER v6: User Manual/Tutorial.
Plymouth: PRIMER-E Ltd.
Althaus, F., Williams, A., Schlacher, T. A., Kloser, R. K., Green, M. A., Barker,
B. A., et al. (2009). Impacts of bottom trawling on deep-coral ecosystems
of seamounts are long-lasting. Mar. Ecol. Prog. Ser. 397, 279–294. doi:
10.3354/meps08248
Anderson, M. J., Gorley, R. N., and Clarke, K. R. (2008). PERMANOVA+ for
PRIMER: Guide to Software and Statistical Methods. Plymouth: PRIMER-E.
Baco, A. R., Rowden, A. A., Levin, L. A., Smith, C. R., and Bowden, D. A. (2010).
Initial characterization of cold seep faunal communities on the New Zealand
Hikurangi margin. Mar. Geol. 272, 251–259. doi: 10.1016/j.margeo.2009.
06.015
Bernardino, A. F., Levin, L. A., Thurber, A. R., and Smith, C. R. (2012).
Comparative composition, diversity and trophic ecology of sediment
macrofauna at vents, seeps and organic falls. PLoS ONE 7:e33515. doi:
10.1371/journal.pone.0033515
Boschen, R. E., Rowden, A. A., Clark, M. R., and Gardner, J. P. A.
(2013). Mining of deep-sea seafloor massive sulphides: a review of the
deposits, their benthic communities, impacts from mining, regulatory
frameworks and management strategies. Ocean Coast. Manage. 84, 54–67. doi:
10.1016/j.ocecoaman.2013.07.005
Boschen, R. E., Rowden, A. A., Clark, M. R., and Gardner, J. P. A. (2015).
Limitations in the use of archived vent mussel samples to assess genetic
connectivity among seafloor massive sulfide deposits: a case study with
implications for environmental management. Front. Mar. Sci. 2:105. doi:
10.3389/fmars.2015.00105
Bowden, D. A., Rowden, A. A., Leduc, D., Beaumont, J., and Clark, M. R. (2016).
Deep-sea seabed habitats: do they support distinct mega-epifaunal benthic
communities that have different vulnerabilities to anthropogenic disturbance?
Deep Sea Res. I 107, 31–47. doi: 10.1016/j.dsr.2015.10.011
Bowden, D. A., Rowden, A. A., Thurber, A. R., Baco, A. R., Levin, L.
A., and Smith, C. R. (2013). Cold seep epifaunal communities on the
Hikurangi margin, New Zealand: composition, succession, and vulnerability
Frontiers in Marine Science | www.frontiersin.org
16
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
Hewitt, J. E., and Thrush, S. F. (2010). Empirical evidence of an approaching
alternate state produced by intrinsic community dynamics, climatic
variability and management actions. Mar. Ecol. Prog. Ser. 413, 267–276.
doi: 10.3354/meps08626
Hewitt, J., Julian, K., and Bone, E. K. (2011). Chatham–Challenger
OceanSurvey20/20 Post-Voyage Analyses: Biotic Habitats and their Sensitivity
to Physical Disturbance. New Zealand Aquatic Environment and Biodiversity
Report. Ministry of Fisheries, Wellington.
Howell, K. L., Davies, J. S., and Narayanaswamy, B. E. (2010a). Identifying
deep sea megafaunal epibenthic assemblages for use in habitat mapping and
marine protected area network design. J. Mar. Biol. Assoc. U.K. 90, 33–68. doi:
10.1017/S0025315409991299
Howell, K. L., Mowles, S. L., and Foggo, A. (2010b).Mounting evidence: near-slope
seamounts are faunally indistinct from an adjacent bank. Mar. Ecol. 31, 52–62.
doi: 10.1111/j.1439-0485.2010.00368.x
Kelly, N. E., Shea, E. K., Metaxas, A., Haedrich, R. L., and Auster, P. J. (2010).
Biodiversity of the deep-sea continental margin bordering the gulf of maine
(NW Atlantic): relationships among sub-regions and to shelf systems. PLoS
ONE 5:e13832. doi: 10.1371/journal.pone.0013832
Leduc, D., Rowden, A. A., Clark, M. R., Bowden, D. A., and Thurber, A. R. (2016).
Limited differences among habitats in deep-sea macro-infaunal communities
off New Zealand: implications for their vulnerability to anthropogenic
disturbance. Mar. Ecol. 37, 845–866. doi: 10.1111/maec.12363
Legendre, P. (1993). Spatial autocorrelation: trouble or new paradigm? Ecology 74,
1659–1673. doi: 10.2307/1939924
Legendre, P., and Anderson, M. J. (1999). Distance-based redundancy
analysis: testing multispecies hypotheses responses in multifactorial
ecological experiments. Ecol. Monogr. 69, 1–24. doi: 10.1890/00129615(1999)069[0001:DBRATM]2.0.CO;2
Legendre, P., and Troussellier, M. (1988). Aquatic heterotrophic bacteria:
modelling in the presence of spatial autocorrelation. Limnol. Oceanogr. 33,
1055–1067. doi: 10.4319/lo.1988.33.5.1055
Levin, L. A., Baco, A. R., Bowden, D., Colaço, A., Cordes, E., Cunha,
M. R., et al. (2016). Hydrothermal vents and methane seeps: rethinking
the sphere of influence. Front. Mar. Sci. 3:72. doi: 10.3389/fmars.2016.
00072
Levin, L. A., Etter, R. J., Rex, M. A., Gooday, A. J., Smith, C. R., Pineda,
J., et al. (2001). Environmental influences on regional deep-sea diversity.
Annu. Rev. Ecol. Syst. 32, 51–93. doi: 10.1146/annurev.ecolsys.32.081501.
114002
Levin, L. A., and Mendoza, G. F. (2007). Community structure and nutrition
of deep methane-seep macrobenthos from the North Pacific (Aleutian)
Margin and the Gulf of Mexico. Mar. Ecol. 28, 131–151. doi: 10.1111/j.14390485.2006.00131.x
Levin, L. A., Ziebis, W., Mendoza, G., Growney, V., Tryon, M., Brown, K., et al.
(2003). Spatial heterogeneity of macrofauna at northern California methane
seeps: the influence of sulfide concentration and fluid flow. Mar. Ecol. Prog.
Ser. 265, 123–139. doi: 10.3354/meps265123
Lundsten, L., Barry, J. P., Cailliet, G. M., Clague, D. A., DeVogelaere, A. P.,
and Geller,J.B. (2009). Benthic invertebrate communities on three seamounts
off southern and central California. Mar. Ecol. Prog. Ser. 374, 23–32. doi:
10.3354/meps07745
Marsh, A. G., Mullineaux, L. S., Young, C. M., and Manahan, D. T.
(2001). Larval dispersal potential of the tubeworm Riftia pachyptila
at deep-sea hydrothermal vents. Nature 411, 77–80. doi: 10.1038/350
75063
Martín, J., Puig, P., Palanques, A., and Ribó, M. (2014). Trawling induced
daily sediment resuspension in the flank of a Mediterranean submarine
canyon. Deep Sea Res. II 104, 174–183. doi: 10.1016/j.dsr2.2013.
05.036
McClain, C. R., Lundsten, L., Ream, M., Barry, J., and DeVogelaere,
A. (2009). Endemicity, biogeography, composition, and community
structure on a Northeast Pacific seamount. PLoS ONE 4:e4141. doi:
10.1371/journal.pone.0004141
Menot, L., Sibuet, M., Carney, R. S., Levin, L. A., Rowe, G. T., Billett, D. S. M.,
et al. (2010). “New perceptions of continental margin biodiversity,” in Life in
the World’s Oceans: Diversity, Distribution, and Abundance, ed A. D McIntyre
(Chichester: Wiley-Blackwell), 79–101.
Clarke, K. R., Somerfield, P. J., and Chapman, M. G. (2006b). On resemblance
measures for ecological studies, including taxonomic dissimilarities and a zeroadjusted Bray–Curtis coefficient for denuded assemblages. J. Exp. Mar. Biol.
Ecol. 330, 55–80. doi: 10.1016/j.jembe.2005.12.017
Clarke, K. R., and Warwick, R. M. (2001). Change in Marine Communities: An
Approach to Statistical Analysis and Interpretation. Plymouth: PRIMER-E Ltd.
Collie, J. S., Escanero, G. A., and Valentine, P. C. (1997). Effects of bottom fishing
on the benthic megafauna of Georges Bank. Mar. Ecol. Prog. Ser. 155, 159–172.
doi: 10.3354/meps155159
Cryer, M., Hartill, B., and O’Shea, S. (2002). Modification of the marine benthos
by trawling: towards a generalization for the deep ocean? Ecol. Appl. 12,
1824–1839. doi: 10.1890/1051-0761(2002)012[1824:MOMBBT]2.0.CO;2
De Leo, F. C., Smith, C. R., Rowden, A. A., Bowden, D. A., and Clark, M. R. (2010).
Submarine canyons: hotspots of benthic biomass and productivity in the deep
sea. Proc. R. Soc. 277, 2783–2792. doi: 10.1098/rspb.2010.0462
De Leo, F. C., Vetter, E. W., Smith, C. R., Rowden, A. A., and McGranaghan,
M. (2014). Spatial scale-dependent habitat heterogeneity influences
submarine canyon macrofaunal abundance and diversity off the Main
and Northwestern Hawaiian Islands. Deep Sea Res. II 104, 267–290. doi:
10.1016/j.dsr2.2013.06.015
de Ronde, C. E. J., Baker, E. T., Massoth, G. J., Lupton, J. E., Wright, I. C., Sparks,
R. J., et al. (2007). Submarine hydrothermal activity along the mid-Kermadec
Arc, New Zealand: large-scale effects on venting. Geochem. Geophys. Geosyst. 8,
Q07007. doi: 10.1029/2006gc001495
Desbruyères, D., Hashimoto, J., and Fabri, M. C. (2006). Composition and
biogeography of hydrothermal vent communities in Western Pacific back-arc
basins. Geophys. Monogr. Ser. 166, 215–234. doi: 10.1029/166gm11
Duineveld, G., Lavaleye, M., Berghuis, E., and Wilde, P. (2001). Activity and
composition of the benthic fauna in the Whittard Canyon and the adjacent
continental slope (NE Atlantic). Oceanol. Acta 24, 69–83. doi: 10.1016/S03991784(00)01129-4
Durden, J. M., Bett, B. J., Jones,. D. O. B., Huvenne,. V. A. I., and Ruhl,.
H. A. (2015). Abyssal hills-hidden source of increased habitat heterogeneity,
benthic megafaunal biomass and diversity in the deep sea. Prog. Oceanogr. 137,
209–218. doi: 10.1016/j.pocean.2015.06.006
Etter, R. J., and Grassle, J. F. (1992). Patterns of species diversity in the deep sea
as a function of sediment particle size diversity. Nature 360, 576–578. doi:
10.1038/360576a0
Fohrmann, M., and Pecher, I. A. (2012). Analysing sand-dominated channel
systems for potential gas-hydrate-reservoirs using an AVO seismic inversion
technique on the Southern Hikurangi Margin, New Zealand. Mar. Petrol. Geol.
38, 19–34. doi: 10.1016/j.marpetgeo.2012.08.001
Genin, A., Dayton, P. K., Lonsdale, P. F., and Spiess, F. N. (1986). Corals
on seamount peaks provide evidence of current acceleration over deep-sea
topography. Nature 322, 59–61. doi: 10.1038/322059a0
Gordon, D. P., Beaumont, J., MacDiarmid, A., Robertson, D. A., and Ahyong, S. T.
(2010). Marine biodiversity of aotearoa New Zealand. PLoS ONE 5:e10905. doi:
10.1371/journal.pone.0010905
Grassle, J. F., Sanders, H. L., Hessler, R. R., Rowe, G. T., and McLellan, T.
(1975). Pattern and zonation: a study of the bathyal megafauna using the
research submersible Alvin. Deep Sea Res. I 22, 457–481. doi: 10.1016/00117471(75)90020-0
Greinert, J., Lewis, K. B., Bialas, J., Pecher, I. A., Rowden, A. A., Bowden,
D. A., et al. (2010). Methane seepage along the Hikurangi Margin, New
Zealand: overview of studies in 2006 and 2007 and new evidence from visual,
bathymetric and hydroacoustic investigations. Mar. Geol. 272, 6–25. doi:
10.1016/j.margeo.2010.01.017
Guennegan, Y., and Martin, V. (1985). “Annexe 3. Techniques de prélèvement,”
in Peuplements Profonds du Golfe de Gascogne, eds L. Laubier and C. Monniot
(Brest: IFREMER), 571–575.
Hanski, I. (1999). Metapopulation Ecology. Oxford: Oxford University Press.
Harris, P. T. (2007). “Application of geophysical information to the design of a
representative system of marine protected areas in southeastern Australia,” in
Mapping the Seafloor for Habitat Characterization, eds B. J. Todd and H. G.
Greene (St. John’s, NL: Geological Association of Canada), 463–481.
Helson, J., Leslie, S., Clement, G., Well, R., and Wood, R. (2010). Private rights,
public benefits: industry-driven seabed protection. Mar. Pol. 34, 557–566. doi:
10.1016/j.marpol.2009.11.002
Frontiers in Marine Science | www.frontiersin.org
17
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
Samadi, S., Schlacher, T. A., and Richer de Forges, B. (2007). “Seamount benthos,”
in Seamounts: Ecology, Fisheries, and Conservation, eds T. J. Pitcher, T. Morato,
P. J. B. Hart, M. R. Clark, N. Haggan, and R. S. Santos (Oxford: Blackwell),
119–140.
Sardà, F., Cartes, J. E., and Company, J. B. (1994). Spatio-temporal variations
in megabenthos abundance in three different habitats of the Catalan deepsea (Western Mediterranean). Mar. Biol. 120, 211–219. doi: 10.1007/BF003
49681
Schlacher, T. A., Schlacher-Hoenlinger, M. A., Williams, A., Althaus, F., Hooper, J.
N. A., and Kloser, R. (2007). Richness and distribution of sponge megabenthos
in continental margin canyons off southeastern Australia. Mar. Ecol. Prog. Ser.
340, 73–88. doi: 10.3354/meps340073
Sellanes, J., Neira, C., Quiroga, E., and Teixido, N. (2010). Diversity patterns
along and across the Chilean margin: a continental slope encompassing oxygen
gradients and methane seep benthic habitats. Mar. Ecol. 31, 111–124. doi:
10.1111/j.1439-0485.2009.00332.x
Sevastou, K., Lampadariou, N., Polymenakou, P. N., and Tselepides, A. (2013).
Benthic communities in the deep Mediterranean Sea: exploring microbial
and meiofaunal patterns in slope and basin ecosystems. Biogeosciences 10,
4861–4878. doi: 10.5194/bg-10-4861-2013
Sibly, R. M., and Hone, J. (2002). Population growth and its determinants: an
overview. Phil. Trans. R. Soc. Lond. B Biol. Sci. 357, 1153–1170. doi: 10.1098/
rstb.2002.1117
Smith, C. R., De Leo, F. C., Bernardino, A. F., Sweetman, A. K., and
Arbizu, P. M., (2008). Abyssal food limitation, ecosystem structure and
climate change. Trends Ecol. Evol. 23, 518–528. doi: 10.1016/j.tree.2008.
05.002
Tecchio, S., Ramírez-Llodra, E., Aguzzi, J., Sanchez-Vidal, A., Flexas, M. M., Sardà,
F., et al. (2013). Seasonal fluctuations of deep megabenthos: finding evidence of
standing stock accumulation in a flux-rich continental slope. Prog. Oceanogr.
118, 188–198. doi: 10.1016/j.pocean.2013.07.015
Tunnicliffe, V., Embley, R. W., Holden, J. F., Butterfield, D. A., Massoth, G. J.,
and Juniper, S. K. (1997). Biological colonization of new hydrothermal vents
following an eruption on Juan de Fuca Ridge. Deep Sea Res. I 44, 1627–1644.
doi: 10.1016/S0967-0637(97)00041-1
Tyler-Walters, H., Rogers, S. I., Marshall, C. E., and Hiscock, K. (2009). A
method to assess the sensitivity of sedimentary communities to fishing
activities. Aquat. Conser. Mar. Freshw. Ecosyst. 19, 285–300. doi: 10.1002/
aqc.965
Van Dover, C. L., Smith, C. R., Ardron, J., Dunn, D., Gjerde, K., Levin, L., et al.
(2012). Designating networks of chemosynthetic ecosystem reserves in the deep
sea. Mar. Policy 36, 378–381. doi: 10.1016/j.marpol.2011.07.002
Vanreusel, A., Fonseca, G., Danovaro, R., Cristina da Silva, M., Esteves, A.
M., Ferrero, T., et al. (2010). The contribution of deep-sea macrohabitat
heterogeneity to global nematode diversity. Mar. Ecol. 31, 6–20. doi:
10.1111/j.1439-0485.2009.00352.x
Vetter, E. W., and Dayton, P. K. (1998). Macrofaunal communities within and
adjacent to a detritus-rich submarine canyon systems. Deep Sea Res. II 45,
25–54. doi: 10.1016/S0967-0645(97)00048-9
Vetter, E. W., Smith, C. R., and De Leo, F. C. (2010). Hawaiian hotspots:
enhanced megafaunal abundance and diversity in submarine canyons on
the oceanic islands of Hawaii. Mar. Ecol. 31, 183–199. doi: 10.1111/j.14390485.2009.00351.x
Watling, L., Guinotte,. J., Clark,. M. R., and Smith,. C. R. (2013). A proposed
biogeography of the deep ocean floor. Prog. Oceanogr. 111, 91–112. doi:
10.1016/j.pocean.2012.11.003
Williams, A., Althaus, F., Dunstan, P. K., Poore, G. C. B., Bax, N. J., Kloser, R. J.,
et al. (2010a). Scales of habitat heterogeneity and megabenthos biodiversity on
an extensive Australian continental margin (100–1100 m depths). Mar. Ecol. 31,
222–236. doi: 10.1111/j.1439-0485.2009.00355.x
Williams, A., Bax, N. J., Kloser, R. J., Althaus, F., Barker, B., and Keith, G.
(2009). Australia’s deep-water network: implications of false homogeneity for
classifying abiotic surrogates of biodiversity. ICES J. Mar. Sci. 66, 214–224. doi:
10.1093/icesjms/fsn189
Williams, A., Schlacher, T. A., Rowden, A. A., Althaus, F., Clark, M. R.,
Bowden, D. A., et al. (2010b). Seamount megabenthic assemblages fail to
recover from trawling impacts. Mar. Ecol. 31, 183–199. doi: 10.1111/j.14390485.2010.00385.x
Metaxas, A. (2011). Spatial patterns of larval abundance at hydrothermal vents
on seamounts: evidence for recruitment limitation. Mar. Ecol. Prog. Ser. 437,
103–117. doi: 10.3354/meps09283
Mountjoy, J. J., Barnes, P. M., and Pettinga, J. R. (2009). Morphostructure
and evolution of submarine canyons across an active margin: cook strait
sector of the Hikurangi Margin, New Zealand. Mar. Geol. 260, 45–68. doi:
10.1016/j.margeo.2009.01.006
Nakajima, R., Yamamoto, H., Kawagucci, S., Takaya, Y., Nozaki, T., Chen, C., et al.
(2015). Post-drilling changes in seabed landscape and megabenthos in a deepsea hydrothermal system, the Iheya North field, Okinawa Trough. PLoS ONE
10:e0123095. doi: 10.1371/journal.pone.0123095
O’Hara, T. D. (2007). Seamounts: centres of endemism or species richness
for Ophiuroids? Glob. Ecol. Biogeogr. 16, 720–732. doi: 10.1111/j.14668238.2007.00329.x
O’Hara, T. D., Rowden, A. A., and Williams, A. (2008). Cold-water coral habitats
on seamounts: do they have a specialist fauna? Divers. Distrib. 14, 925–934. doi:
10.1111/j.1472-4642.2008.00495.x
Palanques, A., Martın, J., Puig, P., Guillen, J., Company, J. B., and Sarda, F.
(2006). Evidence of sediment gravity flows induced by trawling in the Palamos
(Fonera) submarine canyon (northwestern Mediterranean). Deep Sea Res. I 53,
2012–2014. doi: 10.1016/j.dsr.2005.10.003
Palma, M., Quiroga, E., Gallardo, V. A., Arntz, W., Gerdes, D., Schneider,
W., et al. (2005). Macrobenthic animal assemblages of the continental
margin off Chile (22◦ to 42◦ ). J. Mar. Biol. Assoc. U.K. 85, 233–245. doi:
10.1017/S0025315405011112h
Quinn, P. Q., and Keough, M. J. (2009). Experimental Design and Data Analysis for
Biologists. Cambridge: Cambridge University Press.
Ramírez-Llodra, E., Ballesteros, M., Company, J. B., Dantart, L., and Sardà, F.
(2008). Spatio temporal variations of biomass and abundance in bathyal noncrustacean megafauna in the Catalan Sea (North-western Mediterranean). Mar.
Biol. 153, 297–309. doi: 10.1007/s00227-007-0805-y
Ramirez-Llodra, E., Company, J. B., Sarda, F., and Rotllant, G. (2010).
Megabenthic diversity patterns and community structure of the Blanes
submarine canyon and adjacent slope in the Northwestern Mediterranean:
a human overprint? Mar. Ecol. 31, 167–182. doi: 10.1111/j.1439-0485.2009.
00336.x
Ramirez-Llodra, E., Tyler, P. A., Baker, M. C., Bergstad, O. A., Clark, M.
R., Escobar, E., et al. (2011). Man and the last great wilderness: human
impact on the deep sea. PLoS ONE 6:e22588. doi: 10.1371/journal.pone.
0022588
Ramsay, K., Kaiser, M. J., and Hughes, R. N. (1996). Changes in hermit crab feeding
patterns in response to trawling disturbance. Mar. Ecol. Prog. Ser. 144, 63–72.
doi: 10.3354/meps144063
Raymore, P. A. (1982). Photographic investigations on three seamounts in the Gulf
of Alaska. Pac. Sci. 36, 15–34.
Roberts, C. M., Andelman, S., Branch, G., Bustamante, R. H., Castilla, J.
C., Dugan, J., et al. (2003). Ecological criteria for evaluating candidate
sites for marine reserves. Ecol. Appl. 13, S199–S214. doi: 10.1890/10510761(2003)013[0199:ECFECS]2.0.CO;2
Rogers, A. D., Tyler, P. A., Connelly, D. P., Copley, J. T., James, R., Larter, R. D.,
et al. (2012). The discovery of new deep-sea hydrothermal vent communities in
the southern ocean and implications for biogeography. PLoS Biol. 10:e1001234.
doi: 10.1371/journal.pbio.1001234
Rosli, N., Leduc, D., Rowden, A. A., Clark, M. R., Probert, P. K., Berkenbusch,
K., et al. (2016). Differences in meiofauna communities with sediment depth
are greater than habitat effects on the New Zealand continental margin:
implications for vulnerability to anthropogenic disturbance PeerJ 4:e2154. doi:
10.7717/peerj.2154
Rowden, A. A., Schlacher, T. A., Williams, A., Clark, M. R., Stewart,R.,
Althaus, F., et al. (2010a). A test of the seamount oasis hypothesis:
seamounts support higher epibenthic megafaunal biomass than
adjacent slopes. Mar. Ecol. 31, 95–106. doi: 10.1111/j.1439-0485.2010.
00369.x
Rowden, A. A., Schnabel, K. S., Schlacher, T. A., Macpherson, E., Ahyong,
S. T., and Richer de Forges, B. (2010b). Squat lobster assemblages on
seamounts differ from some but not all, deep-sea habitats of comparable
depth. Mar. Ecol. 31(Suppl. 1), 63–83. doi: 10.1111/j.1439-0485.2010.
00374.x
Frontiers in Marine Science | www.frontiersin.org
18
November 2016 | Volume 3 | Article 241
Rowden et al.
Habitat Differences in Deep-Sea Megafaunal Communities
Conflict of Interest Statement: The authors declare that the research was
conducted in the absence of any commercial or financial relationships that could
be construed as a potential conflict of interest.
Wysoczanski, R., and Clark, M. R. (2012). “Southern Kermadec Arc –
Havre Trough geohabitats,” in Seafloor Geomorphology as Benthic Habitat:
GeoHab Atlas of Seafloor Geomorphic Features and Benthic Habitats, eds
P.T. Harris and E. K. Baker (London: Elsevier Insights series. Elsevier),
853–867.
Zajac, R. N., and Whitlach, R. B. (1991). Demographic aspects of marine, soft
sediment patch dynamics. Am. Zool. 31, 808–820. doi: 10.1093/icb/31.6.808
Zeppilli, D., Canals, M., and Danovaro, R. (2012). Pockmarks enhance deep-sea
benthic biodiversity: a case study in the western Mediterranean Sea. Divers.
Distrib. 18, 832–846. doi: 10.1111/j.1472-4642.2011.00859.x
Frontiers in Marine Science | www.frontiersin.org
Copyright © 2016 Rowden, Leduc, Clark and Bowden. This is an open-access article
distributed under the terms of the Creative Commons Attribution License (CC BY).
The use, distribution or reproduction in other forums is permitted, provided the
original author(s) or licensor are credited and that the original publication in this
journal is cited, in accordance with accepted academic practice. No use, distribution
or reproduction is permitted which does not comply with these terms.
19
November 2016 | Volume 3 | Article 241